دستمزدهای منطقه ای و اقدامات ذهنی فرصت های شغلی

We utilise a rich set of regional labour market variables to explain regional variation in Norwegian manufacturing wages. In particular, regional indicators of labour market conditions are computed from survey data in which respondents are asked to evaluate local employment opportunities. We find that average reported satisfaction with local job prospects and other survey-based indicators perform better in regional wage equations than traditional labour market variables, including the regional unemployment rate. Our results suggest that subjective measures of employment opportunities provide useful information about wage pressure.

The present paper shows that ‘subjective’ measures of regional employment opportunities successfully explain regional variations in Norwegian manufacturing wages. The study uses six waves (1993–1998) of an annual survey conducted by the Norwegian Gallup Institute (NGI) in which respondents report how satisfied they are with job prospects in their resident municipality and the surroundings. The survey dataset comprises about 75,000 respondents. Regional measures of employment opportunities computed from the dataset have a positive and significant impact on regional wages, also when controlling for the regional unemployment rate and other ‘objective’ measures of regional labour market tightness. Our results suggest that analyses of wage determination should use subjective measures of labour market conditions in addition to traditional labour market variables. Theories of wage bargaining and efficiency wage theories predict that wages depend on the outside options of workers where outside option depends on labour market pressure. Most empirical studies of wage formation use the unemployment rate as the key labour market tightness indicator.1 The unemployment rate is a proxy for the duration of unemployment and therefore affects the utility loss suffered by workers loosing their jobs. However, the unemployment rate is an imprecise measure of the duration of unemployment. Empirical studies of individual unemployment spells have established that spell duration depends on a number of individual characteristics such as age, education level, previous work experience, family situation and benefit entitlement.2 Because the composition of the unemployment pool varies across the business cycle and between regions, and because the effects of individual characteristics on duration depend on the state of the labour market, there is no stable relation between the unemployment rate and the employment prospects of any given person.3 One possible solution to this problem would be to compute transition rates from unemployment using microdata about individual unemployment spells and use this indicator as proxy for labour market pressure instead of the unemployment rate, cf. Carlsen et al. (2003) for an attempt on this road. In this paper, we propose an alternative set of labour market indicators that provide direct information about the outside options of workers and that can be calculated almost without delay. We utilise information from surveys in which respondents present their beliefs about the state of the regional labour market. The individual responses from the surveys are then aggregated to regional labour market indicators. We can think of several reasons why ‘subjective’ measures of labour market conditions calculated from survey data would explains wages well. First, surveys provide direct information about the beliefs of the insiders, i.e. the agents who determine wages. Subjective measures may therefore reflect characteristics of labour markets considered relevant by insiders but which may not be observable to outside observers. Second, since subjective measures assign weights to different aspects of labour markets according to the beliefs of insiders, changes in the importance of factors relevant to wage determination will automatically be incorporated. Third, insiders sometimes make mistakes. For instance, employers may incorrectly believe that job prospects of a particular type of workers have improved and therefore make generous wage offers to these workers. In such cases, subjective measures, although wrong in an absolute sense, may predict wages accurately. Finally, survey-based indicators can be updated frequently, and surveys can be targeted at labour market segments of particular importance to the wage formation process, e.g. workers and employers involved in early bargaining rounds. Many economists are sceptical to subjective measures of well-being, such as reported satisfaction with various aspects of life. However, in recent years, economists have used subjective measures of well-being to investigate a range of topics, including the costs of unemployment and inflation (Clark and Oswald, 1994, Winkelmann and Winkelmann, 1998, Blanchflower and Oswald, 2000 and Di Tella et al., 2001), the value of direct democracy (Frey and Stutzer, 2000), the importance of absolute versus relative income (Clark and Oswald, 1996), the relation between income and poverty (Ravallion and Lokshin, 2002), the determinants of job separations and quits (Clark, 2001) and the optimal supply of health care resources (Carlsen and Grytten, 2000). These and other contributions have established that responses to questions about well-being are not random numbers, but correlated with relevant objective events and actions. For instance, reported life satisfaction is systematically related to demographic factors, physical and mental health, job loss and family disruption. Job satisfaction is negatively correlated with turnover and absenteeism, satisfaction with community attributes is systematically related to migration decisions, and satisfaction with local public services is systematically correlated with local public spending. The present paper contributes to this literature by showing that subjective assessments of labour markets are systematically related to wage pressure. The paper is organized as follows. Section 2 presents the survey dataset. In Section 3, we investigate the relation between reported satisfaction with job prospects, personal attributes, and objective measures of regional labour market conditions. The results clearly indicate that reported satisfaction is systematically related to relevant personal attributes and several objective measures of regional labour market pressure. In Section 4, we conduct a panel data analysis of regional wages in Norway using objective measures of regional labour market conditions. Section 5 addresses some methodological issues pertaining to aggregation from individual responses to regional variables. In Section 6, we estimate wage equations using both subjective and objective labour market variables. Section 7 offers concluding remarks.

Policy makers collect data about regional labour markets for several reasons. Regional labour market variables are employed to predict wage changes and population movements. Regional labour market variables also provide information about trends in regional disparities. In this paper, we develop a set of regional labour market indicators from surveys in which respondents evaluate local employment opportunities. Our results, based on Norwegian data, suggest a causal relationship between the new indicators and regional wages, and indicate that subjective measures of employment opportunities may improve analyses of wage pressure. We find that average reported satisfaction with local employment opportunities and related subjective measures of regional labour markets perform better than traditional regional labour market variables, including the regional unemployment rate, in regional wage equations. We address a number of methodological issues, and the results turn out to be very robust with respect to the exact specification. These findings suggest that policy makers may want to use subjective measures calculated from surveys as a supplement to traditional sources of information about labour markets. Whether such measures also have a potential for improving analyses of population movements and regional disparities remain issues for future research.